[opensuse] About XFS size in 32/64 bit Linux
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi, I'm preparing an XFS partitioned 2TB hard disk in an external enclosure, using openSUSE 13.1 64 bit, but I intend to use it under 13.1 32 bit. Should I expect problems? I read about an issue regarding xfs and 32 bit vs 64 bit on big partitions, but I'm unsure about what it affects this "early" in the morning :-) What I have read recently is about a weird issue with layout on external enclosures, regarding the sector size of 512/4096. Apparently when accesing via usb the bridge chipset can lie to the kernel or mkfs.xfs so that it thinks that the disk has 512 byte sectors (this does not affect sector start alignment on 1 MB, it's a different issue). Thus, to avoid this possible issue, I want to format the disk using SATA directly, which means using my 64 bit desktop machine, not the older 32 bit machine which doesn't have sata, where the disk is going to be used mostly. (If this question is too specific for this mail list, I'll ask on the xfs mail list instead) - -- Cheers Carlos E. R. (from 13.1 x86_64 "Bottle" at Telcontar) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux) iEYEARECAAYFAlSumaoACgkQtTMYHG2NR9XM5wCgiYrMUMUgYRKjKZBL34bAz1Wz kI4AmwVrknE3S9PXXvnbhHyqk6ViCQ9R =AerN -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 1/8/2015 6:52 AM, Carlos E. R. wrote:
Hi,
I'm preparing an XFS partitioned 2TB hard disk in an external enclosure, using openSUSE 13.1 64 bit, but I intend to use it under 13.1 32 bit.
Should I expect problems?
I read about an issue regarding xfs and 32 bit vs 64 bit on big partitions, but I'm unsure about what it affects this "early" in the morning :-)
What I have read recently is about a weird issue with layout on external enclosures, regarding the sector size of 512/4096. Apparently when accesing via usb the bridge chipset can lie to the kernel or mkfs.xfs so that it thinks that the disk has 512 byte sectors (this does not affect sector start alignment on 1 MB, it's a different issue).
Thus, to avoid this possible issue, I want to format the disk using SATA directly, which means using my 64 bit desktop machine, not the older 32 bit machine which doesn't have sata, where the disk is going to be used mostly.
(If this question is too specific for this mail list, I'll ask on the xfs mail list instead)
-- Cheers Carlos E. R.
(from 13.1 x86_64 "Bottle" at Telcontar)
Not seen anything about this Carlos. I recently installed a much smaller disk (300ish gig) in a Inateck FEU3NS-1 enclosure and installed XFS on it mostly for testing. I had no problem with it, but reformatted the drive for BTRFS mostly as a learning experience, and therefore didn't spend a lot of time using it as XFS. - -- _____________________________________ - ---This space for rent--- -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iEYEARECAAYFAlSu3ZsACgkQv7M3G5+2DLKetACgoRNY0nz3HYBZRp9w3boR3aSR G+cAoIsHlr+Aldw/G/cIO2jvkxopchDs =wPwk -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Donnerstag, 8. Januar 2015 15:52:17 Carlos E. R. wrote:
Hi,
I'm preparing an XFS partitioned 2TB hard disk in an external enclosure, using openSUSE 13.1 64 bit, but I intend to use it under 13.1 32 bit.
Should I expect problems?
I read about an issue regarding xfs and 32 bit vs 64 bit on big partitions, but I'm unsure about what it affects this "early" in the morning :-)
There's an issue of 64 bit inodes on big filesystems, older 32bit userspace, and certain tools. Here's my story: http://oss.sgi.com/archives/xfs/2013-03/msg00088.html XFS is a fairly good filesystem, and will support you nicely even if desaster strikes due to matured recovery tools.
What I have read recently is about a weird issue with layout on external enclosures, regarding the sector size of 512/4096. Apparently when accesing via usb the bridge chipset can lie to the kernel or mkfs.xfs so that it thinks that the disk has 512 byte sectors (this does not affect sector start alignment on 1 MB, it's a different issue).
Thus, to avoid this possible issue, I want to format the disk using SATA directly, which means using my 64 bit desktop machine, not the older 32 bit machine which doesn't have sata, where the disk is going to be used mostly.
I do this all the time with XFS and 4 TB drives and all is doing fine. Cheers, Pete -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 2015-01-08 23:17, Hans-Peter Jansen wrote:
There's an issue of 64 bit inodes on big filesystems, older 32bit userspace, and certain tools. Here's my story:
Curious one, yes... thanks. -- Cheers / Saludos, Carlos E. R. (from 13.1 x86_64 "Bottle" at Telcontar)
On Thursday 08 of January 2015 15:52:17 Carlos E. R. wrote:
Hi,
Hi,
What I have read recently is about a weird issue with layout on external enclosures, regarding the sector size of 512/4096. Apparently when accesing via usb the bridge chipset can lie to the kernel or mkfs.xfs so that it thinks that the disk has 512 byte sectors (this does not affect sector start alignment on 1 MB, it's a different issue).
Thus, to avoid this possible issue, I want to format the disk using SATA directly, which means using my 64 bit desktop machine, not the older 32 bit machine which doesn't have sata, where the disk is going to be used mostly.
I've observed the core issue with two USB to SATA adaptors. The partition table would become mangled if I moved the disk from the internal connector to the adaptor, because the disk was 4kiB/512B and the USB adaptors would report wrong values. -- Regards, Peter -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 2015-01-09 09:29, auxsvr@gmail.com wrote:
On Thursday 08 of January 2015 15:52:17 Carlos E. R. wrote:
I've observed the core issue with two USB to SATA adaptors. The partition table would become mangled if I moved the disk from the internal connector to the adaptor, because the disk was 4kiB/512B and the USB adaptors would report wrong values.
That's a terrible one. The issue that I saw on the XFS mail list is more subtle: "xfs_info says sectsz=512", which is a format value, not runtime. mkfs.xfs gets it from the kerrnel (blockdev), but smartctl/hdparm report the real value. However, it doesn't seem to have much real importance. (thread "What is a recommended XFS sector size for hybrid (512e) advanced format hard drives?") -- Cheers / Saludos, Carlos E. R. (from 13.1 x86_64 "Bottle" at Telcontar)
Carlos E. R. wrote:
On 2015-01-09 09:29, auxsvr@gmail.com wrote:
On Thursday 08 of January 2015 15:52:17 Carlos E. R. wrote:
I've observed the core issue with two USB to SATA adaptors.
That's a terrible one.
(thread "What is a recommended XFS sector size for hybrid (512e) advanced format hard drives?") === It's not just USB adapters. Had an LSI-SAS+SATA RAID card, NEW, a few years back. It didn't recognize the 4KB sector size and reported the devices as having 512-byte sectors (which worked fine if you had all your alignment ducks in a row).
I could format things to 4k, but was a bit paranoid about some FS-layer (partitioner, lvm, mkfs) storing some extra data, "somewhere", such that it might throw off the alignment. At least with 1M partition boundaries and mostly defaults on lvm, mkfs.xfs does allow specifying the 'real' hardware sector size (regardless of what the OS thinks it is -- assuming the real is >= and a multiple of the kernel's idea). Upgrading the card semi-solved the problem -- at least the kernel see the 4096 byte sectors in one place, but other places: Ishtar:/sys/dev/block/8:0/queue> disp_files_w_recursion.sh |grep -P 'sector_size|block_|io_size' hw_sector_size : 512 logical_block_size : 512 minimum_io_size : 4096 optimal_io_size : 0 physical_block_size : 4096 --- It's a bit curious that the 'optimal_io_size' is '0'... does that mean optimally you shouldn't perform i/o on it? ;-) ---- Main thing to watch out for w/32 bit is NOT to enable 64-bit inodes. Once enabled, you can't easily go back -- and I believe it was thought that 32-bit machines could "do" w/32 bits, and if you needed more, then you'd need to move to a 64-bit machine. Not sure if that is still the case, but something to be aware of (and test for if you need it)... -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Freitag, 9. Januar 2015 15:01:50 Linda Walsh wrote:
Main thing to watch out for w/32 bit is NOT to enable 64-bit inodes. Once enabled, you can't easily go back
Actually, you can, the xfs_reno tool allows exactly this. The thread, I've reffered to documents that. Hopefully, the XFS gang have led it bitrod again since then. Cheers, Pete -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
participants (6)
-
auxsvr＠gmail.com -
Carlos E. R. -
Carlos E. R. -
Hans-Peter Jansen -
John Andersen -
Linda Walsh